doi: 10.3389/fmicb.2021.741836
Edited by:
Nuno Pereira Mira, University of Lisbon, Portugal Reviewed by:
Ricardo Jasso-Chávez, Instituto Nacional de Cardiologia Ignacio Chavez, Mexico C. French, University of Edinburgh, United Kingdom
*Correspondence:
Dirk Linke [email protected]
Specialty section:
This article was submitted to Microbial Physiology and Metabolism, a section of the journal Frontiers in Microbiology
Received:15 July 2021 Accepted:03 September 2021 Published:08 October 2021 Citation:
Joudeh N, Saragliadis A, Schulz C, Voigt A, Almaas E and Linke D (2021) Transcriptomic Response Analysis of Escherichia coli to Palladium Stress.
Front. Microbiol. 12:741836.
doi: 10.3389/fmicb.2021.741836
Transcriptomic Response Analysis of Escherichia coli to Palladium Stress
Nadeem Joudeh1, Athanasios Saragliadis1, Christian Schulz2, André Voigt2, Eivind Almaas2and Dirk Linke1*
1Department of Biosciences, University of Oslo, Oslo, Norway,2Department of Biotechnology and Food Science, Norwegian University of Science and Technology (NTNU), Trondheim, Norway
Palladium (Pd), due to its unique catalytic properties, is an industrially important heavy metal especially in the form of nanoparticles. It has a wide range of applications from automobile catalytic converters to the pharmaceutical production of morphine. Bacteria have been used to biologically produce Pd nanoparticles as a new environmentally friendly alternative to the currently used energy-intensive and toxic physicochemical methods. Heavy metals, including Pd, are toxic to bacterial cells and cause general and oxidative stress that hinders the use of bacteria to produce Pd nanoparticles efficiently. In this study, we show in detail the Pd stress-related effects on E. coli.
Pd stress effects were measured as changes in the transcriptome through RNA-Seq after 10 min of exposure to 100 µM sodium tetrachloropalladate (II). We found that 709 out of 3,898 genes were differentially expressed, with 58% of them being up- regulated and 42% of them being down-regulated. Pd was found to induce several common heavy metal stress-related effects but interestingly, Pd causes unique effects too. Our data suggests that Pd disrupts the homeostasis of Fe, Zn, and Cu cellular pools. In addition, the expression of inorganic ion transporters inE. coliwas found to be massively modulated due to Pd intoxication, with 17 out of 31 systems being affected.
Moreover, the expression of several carbohydrate, amino acid, and nucleotide transport and metabolism genes was vastly changed. These results bring us one step closer to the generation of genetically engineeredE. colistrains with enhanced capabilities for Pd nanoparticles synthesis.
Keywords: RNA-seq, heavy metal stress, palladium, precious metals, nanoparticles, oxidative stress,Escherichia coli, transcriptional response
INTRODUCTION
The International Union of Pure and Applied Chemistry (IUPAC) reported in 2002 that there were more than 30 different definitions used in literature for a heavy metal (Duffus, 2002;Pourret, 2018).
Nevertheless, density in most cases is considered to be the defining criterion, and heavy metals are commonly defined by a density higher than 5 g/cm3 (Lozet and Mathieu, 1993;Nies, 1999;
Järup, 2003). Most heavy metals are transition metals with partially filled d orbitals. These d orbitals provide the cations of heavy metals with the ability to form complex compounds which may be redox-active (Nies, 1999). Hence, heavy-metal cations play an important role as trace elements in biological systems.
From a physiological point of view, heavy metals are placed in two categories: (1) essential ones, also known as trace elements, and (2) toxic heavy metals. The essential heavy metals [Co, Cu, Fe, Mn, Mo, Ni, V, W, and Zn (Nies and Silver, 2007)]
are required in smaller amounts than the bulk metals (Na, K, Mg, and Ca). The essential heavy metals have to be acquired by bacteria from their surrounding environments as inorganic ions (Nies and Silver, 2007). These cations are required for functions that include stabilizing biological molecules (Nies and Silver, 2007), electron transfer and redox processes (Bruins et al., 2000), and as components of metalloenzymes, which account for approximately 30% of all enzymes in bacterial cells (Wackett et al., 2004). The toxic heavy metals are elements that have no known beneficial roles and can be damaging to the cell if taken up (Nies and Silver, 2007). The list of toxic heavy metals includes e.g., Ag, Au, Bi, Cd, Cr, Hg, In, Ir, Pb, Pd, Pt, Sn, and Tl (Nies, 1999;Nies and Silver, 2007).
Both excessive levels of essential heavy-metal ions and the presence of toxic heavy-metal ions cause cellular stress (Nies and Silver, 2007). Unlike organic molecules, heavy metals cannot be broken down by enzymatic reactions, and due to the wide range of toxic effects of some heavy metals, strategies for dealing with those toxic heavy metals are limited (Nies and Silver, 2007).
Those strategies include (1) the extracellular detoxification and sequestration of the metal, (2) the prevention of the metal from entering the cell by reducing cell permeability, (3) the active transport of the metal out of the cell, (4) the intracellular sequestration of the metal by binding to proteins, and (5) the enzymatic alteration of the metal to a less toxic form (Rouch et al., 1995).
Bacterial cellular responses to stress caused by metal ions can be measured as changes in the transcription of genes involved in the detoxification processes or metal ion homeostasis. Bacterial transcriptional responses to excessive levels of essential heavy metals have been intensively studied in literature (Kershaw et al., 2005;Lee et al., 2005;Fantino et al., 2010;Gault et al., 2016), but for the toxic heavy metals, this information is often lacking or only exists for key environmental pollutants (Wang and Crowley, 2005;LaVoie and Summers, 2018).
Palladium (Pd), element number 46 in the periodic table, belongs to the platinum group metals and is an industrially important heavy metal due to its unique catalytic properties. It is primarily used as a key component in catalysts for different carbon-carbon coupling reactions such as Heck and Suzuki coupling reactions (Cui et al., 2017; Miller et al., 2017; Biffis et al., 2018). The unique ground-state structure of Pd (4d105s0) and the square-planar geometry of Pd(II) complexes (d8) give Pd unique properties in C-C bond formation and C-O bond cleavage (Jbara et al., 2017). Moreover, Pd is also known for its uniquely high capacity for hydrogen gas absorption (Gavia and Shon, 2015). Besides its major application in automobile catalytic converters (Zereini and Alt, 2006), it is used in versatile applications in catalysis (Biffis et al., 2018), sensors (Kundu et al., 2015), fuel cells (Feliciano-Ramos et al., 2014), and electronics (Zhang G. et al., 2019). Due to its increased industrial use and limited supply, its price has increased more than 27 times in the last 30 years (3,000$/kg in 1990 to 83,000$/kg
in 2021)1. Consequently, the recovery of waste Pd has become of prime importance. The current techniques for Pd recovery include costly and non-environmentally friendly processes, such as pyrometallurgy, solvent extraction, chemical treatment and electrochemical recovery (Baxter-Plant et al., 2002). Therefore, alternative eco-friendly processes for Pd recovery, such as biomineralization, are of high interest (Lloyd et al., 2020).
Pd ions can be taken up and reduced to Pd nanoparticles by living organisms, such as plants (Vishnukumar et al., 2017), fungi (Tarver et al., 2019), and microorganisms (De Corte et al., 2012).
Microorganisms and especially bacteria, due to their fast growth rate and inexpensive cultivation media, are considered one of the most efficient systems for the reduction of Pd (Matsumoto et al., 2020). The formation of Pd nanoparticles through bioreductive deposition of Pd2+ ions was intensively studied in different bacteria (De Corte et al., 2012). This metabolic side process is believed to happen during hydrogen production through anaerobic fermentation in E. coli (Deplanche et al., 2010) and other obligate and facultative anaerobes (Hennebel et al., 2011), or in the process of sulfate and metal reduction in sulfate- reducing (Lloyd et al., 1998) and metal-reducing bacteria (De Windt et al., 2005). Due to its advantageous properties and its status as a model organism,E. colihas been widely used in the recovery of heavy metals and the reduction of metal ions for nanoparticles synthesis (Du et al., 2007;Gurunathan et al., 2009;
Quintelas et al., 2009).
The use ofE. coliwhether for Pd recovery or Pd nanoparticles synthesis requires a deep understanding of the exposure effects that Pd causes to E. coli. In addition, the genetic engineering of E. colistrains that are capable of performing Pd reduction for different applications, requires information about the genes and pathways that are involved in Pd ion uptake from the surrounding environment, in the reduction of Pd ions to form Pd nanoparticles, and in Pd detoxification processes. In this study, we measured the bacterial transcriptional response to the exposure to sublethal levels of Pd2+ions inE. coliK-12 BW25113 strain and compared our results with published transcriptional responses to other heavy metals. Anaerobic conditions were approximated in our experiment since Pd nanoparticles synthesis is typically performed in the absence of oxygen.
MATERIALS AND METHODS
Bacterial Strains and Growth Conditions
The bacterial strain used in this study was E. coli K-12 BW25113 (Baba et al., 2006). For each biological replicate, a single colony of bacteria was used to inoculate a 10 ml starter culture in M63 minimal medium [(NH4)2SO4 2 g/L, KH2PO4 13.6 g/L, and FeSO4.7H2O 0.5 mg/L were mixed together, pH was adjusted to 7 with KOH, the solution was autoclaved and then 1 ml of 1 M sterile MgSO4.7H2O and sterile glucose to a final concentration of 0.4% were added before use] (Miller et al., 1992) incubated overnight at 37◦C with shaking at 200 rpm. On the next day, the overnight starter
1https://www.macrotrends.net/2542/palladium-prices-historical-chart-data
cultures were diluted 1:100 to inoculate 58 ml of the same medium in 50 ml falcon centrifuge tubes and incubated at 37◦C with shaking at 200 rpm until the OD600 reached 0.3.
The falcon centrifuge tubes were filled to the top with 58 ml medium in order to create an anaerobic atmosphere as far as possible. Please note that some residual oxygen availability cannot be excluded as polypropylene tubes are potentially allowing some minimal oxygen diffusion into the sample (Karian, 2003).
A total of 8 samples, 4 palladium-treated and 4 controls, were prepared. Ammonium sulfate, potassium phosphate, ferrous sulfate heptahydrate, potassium hydroxide, magnesium sulfate heptahydrate, and D(+)-Glucose were purchased from Merck Life Science AS, Oslo, Norway.
Determination of Palladium Toxicity
The lethality of Pd to E. coli was measured using the alamarBlueTM Cell Viability Kit (Life Technologies AS, Oslo, Norway). The AlamarBlueTMCell Viability kit uses the reducing power of living cells to quantitatively measure the metabolic activity of bacteria. Resazurin, the active compound of the kit, is a cell permeable compound that is blue in color and non-fluorescent. Upon entering living cells, resazurin is reduced to resorufin, a compound that is red in color and highly fluorescent. Viable cells continuously convert resazurin to resorufin, increasing the overall fluorescence and color of the media surrounding cells.
Two hundred microliter of E. coli bacterial suspension of OD6000.3 in M63 minimal medium were mixed with different Pd concentrations (0.1, 1, and 10 mM) in a 96-well transparent plate. Immediately after that, the cell suspensions were mixed with 10µl of the resazurin reagent (1 mg/ml stock concentration, which corresponds to 4 mM). Three controls with the same concentrations of Pd in 200 µl M63 medium were also mixed with 10 µl resazurin reagent and used for determination of fluorescence background. Then, the fluorescence was measured every 10 min continuously for 8 h at 37◦C using a SYNERGY H1 microplate reader (BioTek, United States). The measurements were fluorescence endpoint with excitation at 530 nm, emission at 590 nm, and 100 gain. The intensity of fluorescence is proportional to the number of living cells and corresponds to the metabolic activity. Fluorescence intensity values (after the subtraction of fluorescence background) were used for the comparison of cell viability between different samples.
Palladium Challenge
When the OD600 reached 0.3, 4 biological replicates were challenged with 100 µM Pd2+ in the form of sodium tetrachloropalladate (II) dissolved in 0.01 M nitric acid (similar amounts of Milli-Q water were added to the 4 control replicates).
The pH of the samples was not affected (data not shown). The cultures were incubated at 37◦C with shaking at 200 rpm for 10 min. The cells were harvested by centrifuging at 4200 × g for 10 min at 4◦C. Then, the pellets were resuspended in 1 ml of 20 mM MOPS buffer (pH 7) and transferred to 2 ml Eppendorf tubes to remove excess heavy metal ions that might impact the subsequent RNA isolation and analysis. The tubes were centrifuged at 6000×gfor 2 min at 4◦C. The supernatant was
removed, and the tubes were immediately dipped into liquid nitrogen, stored at−80◦C, and later shipped on dry ice. Sodium tetrachloropalladate (II) 98%, 5.6 M nitric acid, and MOPS were purchased from Merck Life Science AS, Oslo, Norway.
RNA Sequencing
Total RNA isolation, RNA quality control, library preparation, sequencing, and data analysis were performed by Eurofins Scientific, Konstanz, Germany (INVIEW Transcriptome Bacteria package). Eurofins protocol contained the following steps: total RNA isolation by RNeasy kit (Qiagen, Hilden, Germany), rRNA depletion by NEBNext kit (New England Biolabs, Frankfurt, Germany), and RNA quality measurement by a fragment analyzer (Agilent Technologies, Ratingen, Germany). Then, the mRNA was fragmented and random hexamer primers were used for cDNA synthesis. Adapter ligation and adapter- specific PCR amplification were used to generate libraires of 150 bp reads. More than 10 million reads were generated per sample. The reads were paired-end sequenced by an Illumina sequencing platform.
Data Analysis
The reads were mapped by Eurofins Scientific against the E. coli K-12 MG1655 strain reference genome (the closest E. coli reference genome to E. coli K-12 BW25113 genome) using the software BWA-MEM (version 0.7.12-r1039) (Li, 2013). Transcripts were identified and quantified. Then, pairwise comparison of expression levels and statistical analysis were carried out.
Raw read counts were created using featureCounts (version 1.5.1) (Liao et al., 2014). Only reads overlapping "CDS" features in the reference genome were counted. All reads mapping to features with the same meta-feature attribute were summed. Only reads with unique mapping positions and a mapping quality score of at least 10 were considered for read counting. Supplementary alignments were ignored for read counting. Paired-end reads that mapped with unexpected strandedness were ignored. Reads mapping to multiple features were assigned to the feature that has the largest number of overlapping bases.
A Trimmed Mean of M-values (TMM) normalization was performed using the edgeR package (version 3.16.5) (Robinson et al., 2009;Robinson and Oshlack, 2010). The basic assumption of this normalization technique is that most features (e.g., genes) should not be differentially expressed between samples. For each sample a normalization factor is calculated as the weighted mean of feature-wise log ratios between this sample and a reference sample.
Bioinformatics Analysis
For the interpretation of the results based on gene functions, the differentially expressed genes withp-value<0.05 were manually categorized by Clusters of Orthologous Groups (COGs) using NCBI COGs database2. Next, the differentially expressed genes were furtherly manually subcategorized based on their encoded functions into separate biological processes and pathways by
2https://www.ncbi.nlm.nih.gov/research/cog
Gene Ontology (GO) terms using UniProt3 and Quick GO databases4, and BioCyc pathway/genome database collection5.
Genome-Scale Metabolic Modeling
We applied gene-expression flux balance analysis (gx-FBA) methodology (Navid and Almaas, 2012) to the genome-scale metabolic modelE. coli iML1515 (Monk et al., 2017) in Matlab 2020a (The Math Works Inc., 1991) using Cobra toolbox v3 (Heirendt et al., 2019) with Gurobi (Gurobi Optimization, 2014) as solver. The gx-FBA method integrates gene expression data with the iML1515 model, allowing for response-modeling of environmental perturbations. In short, the cellular metabolic fluxes are scaled in relation to the mRNA expression data in the stress state of the cell. By optimizing for similarity between differentially expressed genes (and hence reaction fluxes) and the provided transcriptomics data, a flux distribution consistent with the constraints and gene expression values of the stressed state is calculated. The flux distribution of the reference state can then be compared with the flux distribution of the stress state. The predicted relative flux changes as a response to Pd stress are used as proxies for metabolic responses.
For modeling the cellular environment corresponding to the biocide environments, we allowed aerobic growth on the default iML1515 medium, using glucose as the sole carbon source. For modeling the cellular environment corresponding to the Pd-stress environment, we set conditions to anaerobic growth and restricted carbon, phosphate and sulfate sources only to those available in the biomass (proteins and glycogen) in minimal amounts necessary to ensure survival, representing the cell using its reserves simply to satisfy ATP maintenance requirements. Secretion fluxes were not changed in the model.
Reactions regulated by non-differentially expressed genes are not directly limited in their flux as they are not part of the gx-FBA objective function, but are nonetheless restricted by substrate availability (and ability for other reactions to consume their products), and therefore, indirectly restricted according to constraints applied to the reaction system. Given a set of constraints, which in our case is imposed by nutrient availability, gene expression, non-depletion/accumulation of metabolites and stoichiometry, an FBA optimal solution (including gx- FBA) will typically not be unique; rather, multiple different flux solutions might satisfy the objective function equally well.
Therefore, flux values for a reaction are not necessarily fixed, but can be said to fall between a pair of upper and lower bounds, with some degree of uncertainty in the flux predictions given by a model.
In each of the gx-FBA simulations, we classify reactions into four categories. The first category, which we call “no flux,”
consists of reactions which do not carry flux in the unstressed environment and are also unable to carry flux in stress conditions.
The second category, “ambiguous,” consists of reactions for which the unstressed reaction flux falls between the upper and lower bounds returned by the gx-FBA optimization, meaning that up-
3https://www.uniprot.org/
4https://www.ebi.ac.uk/QuickGO/
5https://biocyc.org/
or down-regulation of the flux is not a necessary consequence of the gene expression values. The last two categories are down- regulated and up-regulated reaction fluxes. Down-regulated fluxes are reaction fluxes for which the upper bound in the gx-FBA solution is smaller than the unstressed reaction flux, and which must therefore necessarily carry lower reaction flux in stress conditions. Conversely, up-regulated reaction fluxes are those for which the lower bound of the gx-FBA solution are larger than the unstressed reaction flux, which means that the expression values in stress conditions necessarily require an increase in the flux.
RESULTS
Palladium Sublethal Concentration
We have tested several Pd concentrations. Out of the three Pd concentrations used in the viability assay, the highest one (10 mM Pd) resulted in the complete inhibition of metabolism (the green graph inFigure 1). For 1 mM Pd, the cells seemed to be adversely impacted by this concentration, as the fluorescence signal was 22% less than the control in average. After 200 min of incubation at this concentration, the cells went through a phase of adaptation (the shoulder in the blue graph inFigure 1) for ca. 100 min. After that the cells start dying fast. For 0.1 mM Pd, the cells did not appear to be strongly affected, the reduction in fluorescence signal was 7% (the red graph inFigure 1). This concentration was used for Pd exposure in this study.
A Transcriptomic View of Pd Stress
The RNA-Seq results showed that 709 genes, which account for 18.2% of the genes mapped in our experiment, were differentially expressed after 10 min of exposure to 100µM Pd
FIGURE 1 |Effect of Pd on the viability onE. coli.E. colicells were incubated with 10 mM (green triangles), 1 mM (blue triangles), and 0.1 mM (red triangles) sodium tetrachloropalladate (II) or with water as a control (black triangles). The viability of cells was measured for 8 h. The error bars represent the standard deviation of the mean for three replicates.
TABLE 1 |Number of genes that were differentially expressed after 10 min of exposure to 100µM Pd.
Category Number Percentage
Non-differentially expressed genes 3189 81.8%
Differentially expressed genes 709 18.2%
Up-regulated 411 58%
Down-regulated 298 42%
The cut-offp-value was 0.05
From 100% differentially expressed genes 58% were up-regulated and 42%
were down-regulated.
(p-value<0.05). From those genes, 411 (58%) were up-regulated and 298 (42%) were down-regulated. The results are summarized inTable 1.
Differentially Expressed Genes Grouped by Functional Categories
The differentially expressed genes were grouped based on Clusters of Orthologous Groups (COGs) to identify expression differences based on gene functions (Figure 2 and Supplementary Data Sheet 1). Raw RNA-Seq data can be found
inSupplementary Data Sheet 2, Tabs 1–28. In this paper, the COG category names are used together with their international one-letter code (Tatusov et al., 2000).
The top five COG categories with the highest number of differentially expressed genes were amino acid transport and metabolism (D) with 85 genes (50 up-regulated and 35 down- regulated); carbohydrate transport and metabolism (G) with 78 genes (56 up-regulated and 22 down-regulated); transcription (K) with 76 genes (59 up-regulated and 17 down-regulated);
genes of unknown function (S) with 76 genes (49 up-regulated and 27 down-regulated); and post-translational modification, protein turnover, and chaperones (O) with 56 genes (47 up- regulated and 9 down-regulated). The five COG categories with the lowest number of differentially expressed genes were RNA processing and modification (A) with 0 genes affected;
mobilome, prophages, and transposons (X) with 1 up-regulated gene; extracellular structures (W) with 3 down-regulated genes;
intracellular trafficking, secretion, and vesicular transport (U) with 3 genes (2 up-regulated and 1 down-regulated); and cell cycle control, cell division, and chromosome partitioning (D) with 5 genes (3 up-regulated and 2 down-regulated).
The functional categorization of differentially expressed genes showed that there were categories that had a high up-regulation
FIGURE 2 |Counts of differentially expressed genes due to Pd stress grouped by COG functional categories. The cut-offp-value was 0.05. Positive counts represent observed up-regulated genes (green bars) and negative counts represent observed down-regulated genes (red bars). COG code, number of proteins encoded byE. coligenome and category description: (A, 3) RNA processing and modification; (C, 284) Energy production and conversion; (D, 43) Cell cycle control, cell division, and chromosome partitioning; (E, 356) Amino acid transport and metabolism; (F, 107) Nucleotide transport and metabolism; (G, 384) Carbohydrate transport and metabolism; (H, 180) Coenzyme transport and metabolism; (I, 122) Lipid transport and metabolism; (J,236) Translation, ribosomal structure and biogenesis; (K, 301) Transcription; (L, 140) Replication, recombination, and repair; (M, 254) Cell wall, membrane, and envelop biogenesis; (N,110) Cell motility; (O, 178) Post-translational modification, protein turnover, and chaperones; (P, 223) Inorganic ion transport and metabolism; (Q, 72) Secondary metabolites biosynthesis, transport, and catabolism; (R, 261) General function prediction only; (S, 255) Function unknown; (T, 193) Signal transduction mechanisms; (U, 50) Intracellular trafficking, secretion, and vesicular transport; (V, 93) Defense mechanisms; (W, 31) Extracellular structures; and (X, 60) Mobilome, prophages, and transposons.
trend [energy production and conversion (C), transcription (K), and post-translational modification, protein turnover, and chaperones (O)] and other categories with a high down- regulation trend [nucleotide transport and metabolism (F), coenzyme transport and metabolism (H), translation, ribosomal structure and biogenesis (J), and cell motility (N)].
Top Differentially Expressed Genes
The majority of differentially expressed genes (640 out 709 genes) had an absolute fold change (FC) value of 3.5-30 compared to the control (Supplementary Data Sheet 1).Tables 2,3show the top 20 highest up- and down-regulated genes, respectively.
Detailed Results of Differentially
Expressed Genes Based on Clusters of Orthologous Groups Categorization and Gene Ontology Terms
Genes Related to Information Macromolecules DNA replication, recombination, and repair (L)
Of the genes responsible for initiation, maintenance, and termination of chromosome replication, only one gene was down-regulated (recQ) and one was up-regulated (smbC).
Of the genes encoding polymerase subunits, no genes were differentially expressed. Finally, of the genes responsible for DNA repair and recombination, 8 genes were down-regulated and 2 were up-regulated. Detailed information can be found in Supplementary Data Sheet 3, Tab 1.
Transcription (K)
The genes for RNA polymerase subunits were not affected by Pd exposure. The same applies to the termination factors and degradosome complex genes. Of the genes responsible for RNA processing enzymes and RNases, 2 were down- regulated (ghoSandrnb). Three out of the 7 sigma factors were differentially expressed, FecI (ferric citrate sigma factor) was down-regulated, RpoS (stationary phase and stress response sigma factor) and RpoH (heat shock sigma factor) were up-regulated. Detailed information can be found in Supplementary Data Sheet 3, Tab 2. Many transcriptional regulators were differentially expressed due to Pd exposure, 50 were up-regulated and 10 were down-regulated (Supplementary Data Sheet 3, Tab 3). Of all COG categories, transcription had the highest number of up-regulated genes (59 genes).
Translation, ribosomal structure and biogenesis (J)
Six out of the genes responsible for ribosome-assembly proteins (r-proteins) were down-regulated, 5 of them were from the rpsJ operon. Of the other r-proteins, 7 gene were down- regulated (several methyltransferases) and 4 were up-regulated (stress-related proteins responsible for ribosome stabilization).
Moreover, 3 genes responsible for ribosome recycling and stalled- ribosome rescue were up-regulated. The translation initiation and termination factors were unchanged, while 2 of the elongation factors were down-regulated. For the genes of tRNA synthetases/ligases, only one was down-regulated (proline-tRNA ligase). While for tRNA processing genes, 8 were down-regulated and 2 were up-regulated. Detailed information can be found
in Supplementary Data Sheet 3, Tab 4. This shows that the ribosome assembly process and translation overall is repressed due to Pd stress, while at the same time genes for recycling and rescuing present ribosomes are up-regulated.
Post-translational modification, protein turnover, and chaperones (O)
Intoxication with several heavy metals can cause protein cross- linking (Rouch et al., 1995), which disrupts their 3D structures and allosteric movements. The proteases and chaperones of the heat shock response repair or degrade misfolded proteins (Georgopoulos and Welch, 1993). Pd stress induced these repair mechanisms, as 28 different genes for proteases, heat shock proteins, and chaperones were highly up-regulated. Five of them were in the top 20 most up-regulated genes, namely spy (6447-fold),ibpB(565-fold),cpxP(330-fold),ves(151-fold), and ibpA (70-fold). Of the RNases and RNA processing enzyme genes, only one was down-regulated (rnbRNase II, responsible for mRNA degradation). Detailed information can be found in Supplementary Data Sheet 3, Tab 5.
Genes Related to Central Metabolism Carbohydrate transport and metabolism (G)
This COG category had the 2nd highest number of up-regulated genes (56 genes). Most of the up-regulated genes are related to carbohydrate transport. For the transport of disaccharides, malEKX and the regulator gene mlc involved in maltose transporter complex; mglA and mglB involved in galactose transporter complex; and treB and treC involved in trehalose transport and hydrolysis systems, respectively, were up-regulated. For the transport of hexoses,frwA fructose transporter gene; gntP gluconate transporter gene; ytfQRT galactofuranose transporter genes; srlABE sorbitol transporter genes; manXY mannose transporter genes; garP galactarate transporter gene; andL-fucose-proton symporter genefucPwere all up-regulated. The expression of some hexoses metabolic enzymes was also increased. All the genes (fucAIKOR) from L-fucose degradation pathway to make L-lactate and pyruvate;
dgoADKgenes involved in making pyruvate from D-galactonate;
andgarDinvolved in galactarate degradation pathway to make pyruvate were up-regulated.
For the transport of pentoses, xylF xylose transporter gene; rbsA and rbsB involved in the ribose uptake system;
and araF L-arabinose transporter gene were up-regulated.
The sugar efflux transporter gene ydeA responsible for
L-arabinose level maintenance was down-regulated. In addition to transport, several genes from pentose and glucuronate interconversion pathways were up-regulated (uxaABC and uxuAB). For the transport of other kinds of sugars,glpTandglpF glycerol transporter genes and lsrABCDK from autoinducer 2 carbohydrate import system were up-regulated. The function of lsrsystem is not clear, but it has been suggested to have functions in quorum sensing (Xavier and Bassler, 2005).
The other genes that were down-regulated were from unconnected parts of pathways or were poorly defined. Detailed information can be found inSupplementary Data Sheet 4, Tab 1.
The systematic up-regulation of carbohydrate transport genes
TABLE 2 |Top 20 most up-regulated genes after 10 min of exposure to 100µM Pd.
Gene ID Gene name Product description FC Error
b1743 spy Periplasmic ATP-independent protein refolding chaperone 6447 ±2830
b1970 hiuH 5-hydroxyisourate hydrolase, zinc metal-binding 1465 ±986
b3686 ibpB Chaperone, heat shock protein of HSP20 family 565 ±94
b4002 zraP Important component of the zinc-balancing mechanism 357 ±45
b4484 cpxP Part of the cpx two-component envelope stress response system 330 ±92
b4062 soxS Transcriptional activator of the superoxide response regulon 240 ±28
b2074 mdtA Part ofmdtABC-tolC efflux pump, multidrug resistance 188 ±47
b4140 fxsA Suppresses F exclusion of phage T7 163 ±35
b3263 yhdU Function unknown 160 ±148
b1742 ves Cold and stress-inducible protein 151 ±88
b2162 rihB Pyrimidine-specific ribonucleoside hydrolase 133 ±17
b3501 arsR Arsenate resistance operon repressor 132 ±20
b1972 yedZ Protein-methionine-sulfoxide reductase heme-binding subunit MsrQ 112 ±15
b0484 copA Copper-, silver-translocating ATPase efflux pump; involved in copper resistance 101 ±10
b3828 metR Positive activator of the metA, metE and metH genes 100 ±4
b1971 yedY Protein-methionine-sulfoxide reductase catalytic subunit MsrP 96 ±8
b3708 tnaA Tryptophanase, synthesis of pyruvate from L-tryptophan pathway 94 ±3
b4670 yjeV Function unknown 81 ±63
b3502 arsB Arsenite pump; resistance to arsenate, arsenite, and antimonite 76 ±8
b3687 ibpA Chaperone, heat-inducible protein of HSP20 family 70 ±7
TABLE 3 |Top 20 most down-regulated genes after 10 min of exposure to 100µM Pd.
Gene ID Gene name Product description FC Error
b2725 hycA Regulatory protein for the formate hydrogenlyase system −520 ±151
b1038 csgF Curli production assembly/transport component −338 ±60
b3215 yhcA Required for the biogenesis of putative fimbria −197 ±32
b0336 codB Required for cytosine transport into the cell −130 ±23
b2145 yeiS Function unknown −130 ±103
b2269 elaD Protease, capable of cleaving an AMC-ubiquitin model substrate −119 ±13
b4665 ibsC Toxic component of a type I toxin-antitoxin (TA) system −117 ±28
b2345 yfdF Function unknown −117 ±19
b4128 ghoS Antitoxin component of a type V toxin-antitoxin (TA) system −110 ±12
b1520 yneE Predicted inner membrane protein, bestrophin family; possible chloride channel −109 ±27
b0945 pyrD Dihydroorotate dehydrogenase, UMP biosynthesis −106 ±21
b3622 waaL LPS core biosynthesis; O-antigen ligase −85 ±13
b0991 ymcE Cold shock protein, function unknown −80 ±64
b2352 gtrS Serotype-specific glucosyl transferase, CPS-53/KpLE1 prophage −70 ±16
b2497 uraA Uracil permease, uracil transport into the cell −59 ±4
b2669 stpA RNA chaperone and DNA-binding protein −50 ±14
b0564 appY Induces the synthesis of acid phosphatase (AppA) and several other polypeptides −49 ±23
b4345 mcrC McrBC restriction endonuclease −44 ±5
b3508 yhiD Putative magnesium transporter −41 ±16
b1018 efeO Involved in Fe2+uptake −41 ±7
suggests that the cells were in high demand for sugars as they were trying to make energy through different pathways simultaneously to cope with Pd stress.
Energy production and conversion (C)
Several metabolic processes related to energy production were highly modulated upon Pd exposure (shown in detail in Figure 3). For glucose transport and metabolism (Figure 3A),
glucose transport was up-regulated through the up-regulation of one the outer membrane porins, OmpC, and three of the seven glucose transporter systems (Alva et al., 2020).E. colihas three native glycolytic pathways: the Embden-Meyerhof pathway (EMP), the oxidative pentose-phosphate pathway (OPPP), and the Entner-Doudoroff pathway (EDP) (Hollinshead et al., 2016).
None of these pathways were highly modulated, but one enzyme from the EMP, TpiA, was down-regulated. For OPPP,
FIGURE 3 |Palladium (Pd) stress effects on genes related to carbohydrate transport, carbohydrate metabolism and energy production pathways, namely, glucose transport and metabolism(A), glycerol transport and metabolism(B), other sugars transport and metabolism(C), fermentation(D), beta-oxidation cycle(E), aromatic amino acids synthesis(F), and anaerobic respiration(G). In panels(A–F), green gene names represent up-regulated genes, red names represent down-regulated genes, and black names represent unaffected genes. In panel(D), blue color represents fermentation end products. In panel(G), green and red colors represent up-regulated and down-regulated parts of anaerobic respiration pathways, respectively. In panel(A), the three native glycolytic pathways inE. coli are demonstrated: Embden-Meyerhof pathway (EMP), oxidative pentose-phosphate pathway (OPPP), and Entner-Doudoroff pathway (EDP). The full names of abbreviated compounds are DHAP, dihydroxyacetone phosphate; KDPG, 2-Keto-3-deoxy-6-phosphogluconate; DHA, dihydroxyacetone; KDG,
2-dehydro-3-deoxy-D-gluconate; TCA, tricarboxylic acid cycle; DMSO, dimethyl sulfoxide; and TMAO, Trimethylamine N-oxide.
2 enzymes were down-regulated, TalA and PykF, and one was up-regulated, RpiB. Interestingly, the expression of the genes responsible for the transport and metabolism of glycerol (Figure 3B) and other sugars (Figure 3C) was highly increased despite the fact that these sugars were not present in the culture medium. Glycerol transport and metabolism genes had a high up-regulated trend; its two transport systems, GlpQ enzyme (glycerophosphoryl diester phosphodiesterase which hydrolyzes deacylated phospholipids to glycerol-3-phosphate), and several of its metabolic enzymes were up-regulated. For the transport and metabolism of other sugars, five sugar degradation pathways to make pyruvate were up-regulated (from
L-fucose, D-glucuronate, D-galacturonate, D-galactonate, and D-galactarate) together with the transporters of L-fucose and D-galactarate.
Fermentation processes (Figure 3D) were also modulated.
The production of D-lactate, malate, and fumarate was increase thought the up-regulation ofldhA,mdh, andfumC, respectively.
The production of H2was highly modulated thought the massive down-regulation of the transcriptional repressor of hydrogenase 3 operon, HycA, which was the most down-regulated gene due to Pd exposure, decreasing 520-fold. Moreover, two other pathways that are related to fermentation were highly modulated. The beta- oxidation cycle (Figure 3E) that produces acetyl-CoA was highly up-regulated through the up-regulation of the outer membrane
protein FadL and the inner membrane protein FadD responsible for the transport of saturated and unsaturated fatty acids, together with the beta-oxidation cycle genes,fadABEHIJ.
Phosphoenolpyruvate enters two separate pathways; it is either converted to oxaloacetate or to pyruvate. In addition, phosphoenolpyruvate is also the precursor for aromatic amino acids synthesis (Figure 3F) together with erythrose-4-P. This pathway was massively down-regulated on two levels, firstly, through the down-regulation ofaroAEFGHgenes which encode enzymes that convert phosphoenolpyruvate and erythrose-4-P to chorismate, and secondly, through the down-regulation of aromatic amino acids individual pathways, throughtyrA,pheA, andtrpEgenes.
Besides the modulation of glycolysis and fermentation processes, the anaerobic respiration process was also modulated. In this study, HNO3 was used a solvent for sodium tetrachloropalladate (II). The addition of HNO3 (0.291 µL from 0.01 M stock concentration to 58 mL medium) created a 50µM HNO3 concentration in the medium. E. colican use the nitrate of HNO3as terminal electron acceptor for anaerobic respiration. The nitrate reductase operons ofE. coliare optimally expressed at higher nitrate concentrations. Genes such asnapF, nrfA, and nirBare optimally expressed at 1 mM nitrate while narG is optimally expressed at 10 mM nitrate concentration (Wang et al., 1999; Wang and Gunsalus, 2000). However, the
minimal HNO3concentration added might be the cause for this modulation, at least partially (Figure 3G).
For respiration inE. coli,dehydrogenases transfer electrons from various substrates to the quinone pool. Then, terminal reductases and oxidases transfer these electrons to different electron acceptors. For aerobic respiration, the terminal electron acceptor is O2, while for anaerobic respiration there are different electron acceptors. On the dehydrogenases level, the expression of formate and glycerol-3-P dehydrogenases and the expression of hydrogenase 2 was increased, while the expression of malate dehydrogenase was decreased. On the reductases level, the expression of nitrate and methionine sulfoxide reductases was up-regulated.
In addition to all this, the expression of the glucose dehydrogenase gene gcd and 4 of the 6 genes of the rsx operon (ion-translocating oxidoreductase complex) were down-regulated. This complex is a membrane-bound complex that couples electron transfer with translocation of protons across the membrane. Detailed information about energy production and conversion modulated genes can be found in Supplementary Data Sheet 4, Tab 2.
Amino acid transport and metabolism (E)
This category of COGs was modulated the most under Pd stress with 85 genes being differentially expressed, 50 up-regulated and
35 down-regulated. We found particular amino acid pathways that were heavily modulated (more than 5 genes of the pathway), others that were slightly modulated with 2-3 genes being affected, and the rest had only one or no genes being modulated. The amino acid synthesis pathways that were heavily modulated are shown in Figure 4. The modulation is divided into two parts: biosynthesis pathways and uptake pathways. On the biosynthesis level, arginine biosynthesis was highly up-regulated, with all the 8 genes involved in its biosynthesis from glutamate (argABCDEGHI) were up-regulated. Arginine could alternatively be synthesized from glutamine by converting glutamine to carbamoyl phosphate by the products of the carA and carB genes; but we found this pathway was down-regulated. Glutamate together with putrescine can be also converted to succinate by the products ofpuuABCDEgenes, this pathway was dramatically down-regulated. It appears that bacteria are saving all the glutamate for arginine biosynthesis, and particularly through this pathway and not from glutamine (Figure 4A).
The biosynthesis of histidine (Figure 4B) and methionine (Figure 4C) was also highly up-regulated, where the genes hisABCDGH and metABCFL were up-regulated, respectively.
Moreover, the expression of the positive regulator of the methionine operon, MetR, increased 100-fold. At the same time, the biosynthesis of aromatic amino acids was severely down-regulated. For their biosynthesis, phosphoenolpyruvate
FIGURE 4 |Amino acid-related pathways that were heavily modulated due to Pd stress. In panels(A–D), green gene names represent up-regulated genes, red names represent down-regulated genes, and black names represent unaffected genes. In figure(E), green, red, and black letters represent up-regulated, down-regulated, and unaffected genes, respectively.(A), arginine biosynthesis from glutamate (argABCDEGHI) was highly up-regulated, but the biosynthesis of arginine from glutamine was down-regulated (carAandcarBgenes). The conversion of glutamate and putrescine to succinate by the productspuuABCDEgenes was also heavily down-regulated. The biosynthesis of histidine(B)and methionine(C)was also highly up-regulated.(D), the biosynthesis of aromatic amino acids was drastically down-regulated on two levels, firstly, thought the down-regulation of the production of their common precursor (chorismate) from
erythrose-4-phosphate by the products ofaroABCDEFGHgenes, and secondly, through the down-regulation of their individual pathways.(E), the pathways involved in the transport of glutamine, arginine, histidine, and methionine were up-regulated. Lastly, the putrescine importer genepuuPwas down-regulated.
and erythrose-4-P have to be converted to chorismate by the products of aroABCDEFGH genes, then, chorismate can be converted to the three aromatic amino acids through three different pathways. For tyrosine and phenylalanine, the first step in their biosynthesis is that chorismate is converted to prephenate by the products of tyrA and pheA genes, respectively, while for tryptophan, chorismate is converted to anthranilate by the product of trpE gene. The majority of all these genes were down-regulated (Figure 4D).
On the uptake level (Figure 4E), a comparable behavior was found. The pathways involved in the transport of glutamine, arginine, histidine, and methionine were up-regulated as well. For glutamine and methionine, all the three genes involved in their transporter complex were up-regulated (the periplasmic binding protein (GlnH and MetQ, respectively), the membrane permease (GlnP and MetI, respectively), and the ATP-binding import protein [GlnQ and MetN, respectively)]. For histidine, the genes for the periplasmic binding protein HisJ and the permease HisQ were up-regulated. For arginine, only the periplasmic binding protein ArgI was up-regulated. Lastly, the putrescine importer PuuP was down-regulated.
For the amino acid pathways that were slightly modulated, 2 genes involved in the serine biosynthesis pathway were down-regulated, serA and serC. Two genes involved in the isoleucine biosynthesis pathway were down-regulated, ilvC andilvD. The three genes involved in the transporter complex
of glycine betaine, yehWXY, were down-regulated. The genes encoding the S-methylmethionine permease, mmuP, and homocysteine S-methyltransferase, mmuM, were up-regulated.
Lastly, two genes involved in the catabolism of tryptophan,tnaA andtnaB, were up-regulated. Detailed information can be found inSupplementary Data Sheet 5, Tab 1.
Inorganic ion transport and metallochaperones (P)
Escherichia coli has several systems for importing, balancing, and exporting bulk and trace metal ions. Some of these systems are very specific to one metal ion while others can transport a variety of ions. Figure 5 shows the well-studied systems in E. coli, the upper half of the figure shows the systems responsible for ion import while the lower half shows the systems responsible for ion export. Ten of the 17 systems responsible for importing inorganic ions were differentially expressed, 8 down-regulated and 2 up-regulated. The periplasmic binding protein gene nikA of the nickel transporter complex operon nikABCDE was down-regulated. The uptake protein genetrkG of the potassium transporter complex operontrkAEGwas down- regulated. Interestingly, all the 6 systems responsible for iron import were down-regulated, namely, the ferric citrate sigma factor gene fecI and the regulator gene fecR responsible for the ficABCDE operon of the Fe3+-citrate import system; the periplasmic binding protein gene feoA and the GTP-binding channel protein genefeoBoffeoABCoperon of the Fe2+import
FIGURE 5 |Palladium (Pd) stress effects on inorganic metal ion import and export systems inE. coli. The upper half of the figure shows the systems responsible for ion import and the lower half shows the systems responsible for ion export. Some systems are specific to one metal ion while other can transport a variety of ions.
The small rectangles in the inner membrane indicate that the system has an inner member permease protein and the rectangles in the outer membrane indicate that the system includes an outer membrane porin. The up-regulated genes due to Pd stress are shown in green, the down-regulated genes in red, and the unaffected genes in black. Ten of the importing systems were modulated, 8 of them were down-regulated. All the iron importing systems were down-regulated. Seven of the exporting systems were modulated, 6 of them were up-regulated.
system; the periplasmic binding protein gene efeO and the deferrochelatase gene efeB of theefeUOB operon of the Fe2+ import system; the ATP-binding protein genefebCoffepABCDG operon of the Fe3+-enterobactin import system; the siderophore receptor gene fiu responsible for iron transport across the outer membrane complexed with catecholate siderophores such as dihydroxybenzoylserine and dihydroxybenzoate; and the permease protein gene fhuB of fhuBCD operon of the Fe3+- hydroxamate import system. The up-regulated genes were the magnesium-transporting ATPase gene mgtA and periplasmic zinc chaperon genezinT.
Seven of the 14 systems responsible for exporting inorganic ions were differentially expressed, 6 up-regulated and 1 down-regulated. The up-regulated genes werercnAandrcnBof the nickel/cobalt efflux system; the transporting ATPase gene zntA and the regulator gene zntR of the zinc/cadmium/lead system; the copper-exporting ATPase gene copA; the channel protein genearsB, arsenate reductase genearsC, and the regulator gene arsR of the arsenical resistance system; the sodium- potassium/proton antiporter genechaA; and the sodium/proton antiporter gene nhaA together with its regulator gene nhaR.
The down-regulated genes were the glutathione-regulated potassium-efflux system genekefCand its regulator genekefF.
In addition to the transporter complexes, several other metal-related pathways were modulated as well. The genes involved in the biosynthesis of the enterobactin siderophore, a low-molecular mass compound that is secreted extracellularly byE. colito chelate Fe3+ from the environment and then it is brought back to the cell through the FepABCDG system (Gehring et al., 1997), were down-regulated (entC and entE). The iron storage system was up-regulated throughftnBgene. All the genes of theiscARSU operon involved in Fe-S cluster assembly were up-regulated. The gene for copper oxidase, cueO, was highly up-regulated, increasing 44-fold.
Three other non-metal transporters were also modulated.
The phosphate transporter complex permease gene pstA was down-regulated, the nitrate/nitrite transporter genes narK and narUwere down-regulated, and the periplasmic binding protein gene tauA and the ATP-binding protein gene tauB of taurine import system were up-regulated. Detailed information can be found inSupplementary Data Sheet 5, Tab 2.
Coenzyme transport and metabolism (H)
This COG category had a high down-regulation trend, mainly on the vitamin biosynthesis level. Nine of the 11 genes involved in vitamin B1 (thiamine) biosynthesis were down-regulated (thiCDEFGHIMS). Two genes involved in vitamin B2 (riboflavin) biosynthesis pathway were down-regulated (ribAandyigB). For vitamin B3 (nicotinamide riboside), its transporter genepnuC, and one gene involved in its biosynthesis (pncB) were down- regulated. Of the vitamin B6 (pyridoxal phosphate) biosynthesis pathway, 2 genes were down-regulated (pdxKandpdxY). Of the vitamin B9 (folate) biosynthesis pathway, 2 gene were down- regulated (folEandfolK). For vitamin B12 (cobalamin), one gene of its transporter complex to the cell was down-regulated (the ATP-binding protein genebtuD), but one gene of its biosynthesis pathway was up-regulated (btuR).
In addition to vitamins, the synthesis of quinones was modulated. Two genes involved in ubiquinone biosynthesis pathway were up-regulated (ubiA and ubiC), but 2 genes involved in menaquinone/demethyl-menaquinone biosynthesis pathway were down-regulated (menF andmenH).
Moreover, the biosynthesis of iminosuccinic acid through the oxidation of aspartate was down-regulated through nadB. Phenylacetate degradation pathway was up-regulated through paaJ andpaaK. Detailed information can be found in Supplementary Data Sheet 5, Tab 3.
Nucleotide transport and metabolism (F)
This COG category had a high down-regulation trend too, 28 genes were down-regulated (3 of them were in top 20 most down-regulated genes) compared to only 4 up-regulated genes. For purine biosynthesis pathway from phosphoribosyl pyrophosphate (PRPP), 9 genes were down-regulated, purBCEFHKMNU. For pyrimidine biosynthesis pathway from bicarbonate, 4 genes were down-regulated, pyrBCDI. For uridine monophosphate (UMP) biosynthesis pathway, 2 genes were down-regulated, carA and carB. Finally, for guanosine monophosphate (GMP) biosynthesis pathway, 2 genes were down-regulated, guaA and guaB. Of purine and pyrimidine hydrolysis pathways, 3 genes were drastically up-regulated,rihB, rihCandhiuH, the first being the 11th and last being the 2nd highest up-regulated genes with 133- and 1465-fold, respectively.
The genes involved for the biosynthesis of nucleotides through the nucleotide salvage pathway were also down-regulated: for adenosine monophosphate (AMP), apt was down-regulated;
for cytidine monophosphate (CMP),codAwas down-regulated;
for GMP and xanthosine monophosphate (XMP), gpt was down-regulated; for inosine monophosphate (IMP), gsk was down-regulated; and for UMP,uppwas down-regulated.
For the transport of nucleotides, all the permeases involved in the uptake of nucleobases were down-regulated.
Adenine permease gene adeP, cytosine permease gene codB, guanine/hypoxanthine permease gene ghxP, uracil permease gene uraA, xanthine permease gene xanP, and the nucleoside permease gene nupC. Detailed information can be found in Supplementary Data Sheet 5, Tab 4.
Lipid transport and metabolism (I)
This category had a low number of modulated genes, 14 up-regulated and 7 down-regulated. The key pathway that was affected was the fatty acid beta-oxidation cycle, where 6 of its genes, fadABEHIJ, were up-regulated. The fatty acid transport protein genesfadLandfadDwere also up-regulated. Two genes of isopentenyl diphosphate biosynthesis via DXP pathway,ispDand ispF, were down-regulated. Detailed information can be found in Supplementary Data Sheet 5, Tab 5.
Cell Cycle Control, Cell Division, and Chromosome Partitioning (D)
For this category, only a few genes were differentially expressed.
The cell death peptidase gene lit and the cell division protein gene yedR were down-regulated. The toxic component gene ibsC of type I toxin-antitoxin (TA) system (Fozo et al., 2008) and the antitoxin component gene ghoSof type V TA system
(Cheng et al., 2014) were heavily down-regulated, being the 7th and 9th most down-regulated, decreasing 117- and 110-fold, respectively. The toxic component genemokCof the type I TA system (Pedersen and Gerdes, 1999) was up-regulated.
Cell Wall Functions
Cell wall biogenesis, defense mechanisms, and extracellular structures (M, V, and W)
For cell wall biogenesis, several genes involved in LPS biosynthesis were down-regulated,arnC,eptC,IptD,lpxH,waaC, waaLandwza. Several genes involved in enterobacterial common antigen biosynthesis pathway were also down-regulated, rffG, rffH, wecBandwecC. Moreover,osmB gene of the osmotically inducible lipoprotein B, which provides resistance to osmotic stress, was up-regulated.
More importantly,E. coli encodes multiple antibiotic efflux systems that increase drug efflux and limit passive uptake by decreasing porin expression (Alekshun and Levy, 1999). These genes are activated in response to antibiotics and general stress.
We found that Pd stress activates them too. From the multiple antibiotic resistance MarRAB system, 3 out of 4 genes were highly up-regulated,marABR. Several TolC-dependent systems were modulated too. From the multidrug resistance MdtABC system, 5 genes were vastly up-regulated,mdtABCDG,mdtAwas the 7th highest up-regulated gene, increasing 188-fold. While for the anaerobic multidrug efflux transporter genes mdtEF, mdtE was down-regulated. From the AcrAB system, 2 genes were up- regulated,arcD andarcR. From the EmrAB energy-dependent system,emrB was down-regulated. This might suggest thatE. coli is trying to use the passive efflux systems and not the active ones, to save energy for other crucial systems. Finally, from the outer membrane porins (Omp) system,ompCwas up-regulated.
OmpC imports extracellular glucose to the periplasmic space (Alva et al., 2020), the high need for energy is possibly the reason
for this up-regulation. Detailed information can be found in Supplementary Data Sheet 6, Tab 1.
Motility and biofilm (N)
This category had a heavy down-regulation trend with a total of 16 genes being down-regulated compared to only 3 up-regulated genes. Of the flg operon responsible for flagellar synthesis, 8 genes, flgBCDEFGHI, were down-regulated. The same effect was found on type 1 fimbriae, which are filamentous pili that are attached to the cell surface and mediate biofilm formation and adhesion onto host cells. FimBCDFG, encoding proteins responsible for the assembly of FimA fimbrial subunit, its export across the outer membrane, and its regulation, were all down-regulated. Moreover, ecpB was also down-regulated, which is a member of ecpRABCDE operon that encodes the E. coli common pilus (ECP) and plays a role in early-stage biofilm development and host cell recognition. In addition, genes involved in curli organelles biogenesis were heavily down- regulated, the expression ofcsgEandcsgFwas 16- and 338-fold lower, respectively. At the same time, the 3 genes that were up- regulated were bssS(a biofilm repressor gene),tomB(a biofilm formation regulator gene), andglgS(a negative regulator gene of motility, adhesion, and synthesis of biofilm exopolysaccharides).
In summary, this all shows that the bacteria were intentionally avoiding the structurally and energetically intensive motility and biofilm formation systems. Detailed information can be found in Supplementary Data Sheet 6, Tab 2.
Stress Responses
Oxidative stress response and repair
There are two oxidative stress response systems in E. coli, the OxyRS and SoxRS regulons (Seo et al., 2015), explained in Figure 6. OxyR is a member of the LysR transcriptional regulators family, which uses a cysteine-pair to sense the oxidative damage
FIGURE 6 |The oxidative stress response systems inE. coli.
and regulates 49 genes when oxidized (Choi et al., 2001).
Nine genes out of those 49 were up-regulated due to Pd stress and no genes were down-regulated. oxySis a small non- coding RNA (ncRNA) regulated by OxyR, which repressesrpoS (stationary phase and stress response sigma factor gene),fhlABC flagellar proteins genes, and other genes to prevent redundant induction of stress response genes (Altuvia et al., 1998). Our RNA purification and library preparation methods were not optimized for the enrichment of nc-RNAs; therefore, we were not able to detect any of them. Despite being unable to detect the expression ofoxyS, several genes regulated by it were modulated.
The expression of flhA and flhC was down-regulated but the expression ofrpoSwas up-regulated. The up-regulation ofrpoS might be a due to the low number of OxyR stress response activated genes (9 out of 49), which resulted in no redundant induction of stress response genes.
The SoxRS regulon is the other oxidative stress response system in E. coli. SoxR is a member of the MerR repressor-activator family, which uses the oxidation state of 2Fe-2S clusters to sense the oxidative damage and regulates 10 stress-related genes (Watanabe et al., 2008). The most important gene of SoxR regulated genes is soxS, which is responsible for the regulation of additional 53 genes (Seo et al., 2015). Out of the 10 genes regulated by SoxR, 4 genes were up-regulated and 3 were down-regulated. soxSwas modulated the highest in this list, its expression was up-regulated 240-fold. Then, of the 53 genes regulated by SoxS, 15 genes were up-regulated and 4 were down-regulated. This altogether shows that Pd exposure caused a serious oxidative stress toE. coli. Detailed information can be found inSupplementary Data Sheet 7.
Genome-Scale Metabolic Modeling
First, we note that 306 of the 709 differentially expressed genes are present in the genome-scale metabolic model iML1515 (Monk et al., 2017). In Figures 7, 8, we present the E. coli K-12 MG1655 KEGG (Kyoto Encyclopedia of Genes and Genomes) (Kanehisa, 2019; Kanehisa et al., 2019) pathway map. Colored in green and red are the up- and down-regulated reaction flux predictions based on the iML1515 genome- scale metabolic model, respectively, with yellow representing non-zero reaction fluxes that are not required to be regulated.
InTable 4we present the individual regulation pattern for ten different biocides (Pereira et al., 2020) as well as for Pd. We found that the general distribution of reactions into different categories is fairly consistent between biocides, with between 50 and 125 up-regulated reactions, 550 to 600 down-regulated reactions, and eight to 15 ambiguous reactions. There are approximately 2,000 reactions in the metabolic network that do not carry flux. The Pd flux distribution differs substantially from the others, with the majority of reactions (1,485 out of 2,712) being categorized as ambiguous, 1,090 not carrying flux, 17 up-regulated, and 120 down-regulated reactions.
Comparing the stress responses between the Pd stress environment and the biocide set as a whole (using the median flux values and upper and lower bounds of each reaction across biocides as a proxy for a generic response), we found the most significant commonality between the down-regulated
sets (Table 5). Almost all reactions down-regulated in Pd (105 out of 120) are also being generically down-regulated under biocide stress conditions. However, we also found several major differences: in particular, ambiguous reactions are far more common in the Pd set, representing more than half (1,485 out of 2,712, or 55%) of all of the reactions. This is also reflected in Figures 7, 8, where we can see that the Pd condition is dominated by ambiguous fluxes, whereas biocide stress in general corresponds to consistently down-regulated fluxes.
DISCUSSION
The general and oxidative stress-related effects of Pd stress to E. coliwere compared to published work on the transcriptomic response ofE. colito other heavy metal stresses, namely, Ni (Gault et al., 2016), Co (Fantino et al., 2010), Cu (Kershaw et al., 2005), Zn (Lee et al., 2005), Ag (McQuillan and Shaw, 2014), Hg (LaVoie and Summers, 2018), and Cd (Helbig et al., 2008; A.Wang and Crowley, 2005), and to the response of the Enterobacteriaceae bacterium LSJC7 to As stress (Zhang et al., 2016). Pd-induced oxidative stress effects were also compared to other non-heavy metal-related oxidative stresses, namely, Cl (Wang et al., 2009) and biocide (Pereira et al., 2020) stresses. We found that several Pd stress-related effects are similar to the effects of other heavy metal or oxidative stresses (discussed in the section on common transcriptional changes, below). Interestingly, Pd stress caused unique effects that were not reported in the aforementioned heavy metal and oxidative stress studies (discussed in uncommon transcriptional changes, below).
Common Transcriptional Changes
This section discusses transcriptional changes that have been shown to occur in heavy metal stress responses previously.
Overall, the challenge with Pd ions induced a multitude of expected changes in general stress response mechanisms and especially in efflux systems. A number of changes in basic energy metabolism were also observed in other heavy metal stress responses previously.
Protein Synthesis Arrest
Protein synthesis uses most of the cellular ATP (Pontes et al., 2015). The global down-regulation of the translation apparatus including the expression of the r-proteins, the elongation factors, and the tRNA processing genes suggests that the cells are trying to save energy for stress-related pathways rather than translation.
This process of energy conservation was also found in other heavy metal stresses including Ni (Gault et al., 2016), Hg (LaVoie and Summers, 2018), Ag (McQuillan and Shaw, 2014) and Cd (Wang and Crowley, 2005).
RpoH and RpoS Sigma Factor Up-Regulation
When E. coli is exposed to high temperatures or any other condition that results in unfolded proteins, including exposure to heavy metals, the heat shock sigma factor, RpoH, is induced (Foster, 2005). RpoH regulates the expression of approximately 30 genes encoding heat shock response proteins. In contrast to
FIGURE 7 |Predicted up- and down-regulated median fluxes across ten biocides (Pereira et al., 2020) using theiML1515 model, presented in the KEGG pathway map ofE. coli. Up-regulated fluxes are green, down regulated fluxes red, and fluxes without a regulation requirement are colored yellow. No-flux reactions and reactions without KEGG IDs are omitted.
FIGURE 8 |Predicted up- and down-regulated fluxes in response to Pd stress using theiML1515 genome-scale metabolic model, presented for the KEGG pathway map ofE. coli. Up-regulated fluxes are green, down regulated fluxes red, and fluxes without a regulation requirement are colored yellow. No-flux reactions and reactions without KEGG IDs are omitted.
